Scholarlink

Efficient water planning crucially relies on monitoring surface water resources. In this context, threshold-based index methods are a less complex yet accurate approach to separate waterbodies from satellite imageries. Therefore, this study aims to identify efficient water mapping methods by comparing the performance of various water indices on Sentinel-2 scenes in the Eastern Hills (EH) of Bangladesh, using it as a case study. Here, the popular indices NDWI, MNDWI, NDMI, WRI, AWEInsh, AWEIsh, NWI, and SWI, are assessed. While finding the best panchromatic band, SWIR1 has been found to have a strong association (R2 = 0.512) with the average of four bands: Red, Green, Blue, and NIR. Similarly, VNIR and SWIR2 are found to be strongly correlated with the Red Band with R2 values of 0.68 and 0.43, respectively. High Pass Filter has the highest Universal Quality Index, lowest Spectral Angle Mapper, and lowest Root Mean Squared Error for pan-sharpening SWIR1, SWIR2, and VNIR bands. NDWI shows the highest Kappa (0.87) and F1-score (0.92). Following historical (2016–2023) analysis, Yen thresholding method shows an overall accuracy ranging from 82% in 2023 to 96% in 2021. Thus, the study suggests using Yen thresholding for NDWI in EH of Bangladesh.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Assessing Spatial Thresholds of Indices-Based Water Mapping with Sentinel-2 for Ukhia and Teknaf of Bangladesh

  • 章节

  • 发表日期：

    2025年

查看全文
  • Saifullah Sayed,
  • Sara Nowreen

摘要

Efficient water planning crucially relies on monitoring surface water resources. In this context, threshold-based index methods are a less complex yet accurate approach to separate waterbodies from satellite imageries. Therefore, this study aims to identify efficient water mapping methods by comparing the performance of various water indices on Sentinel-2 scenes in the Eastern Hills (EH) of Bangladesh, using it as a case study. Here, the popular indices NDWI, MNDWI, NDMI, WRI, AWEInsh, AWEIsh, NWI, and SWI, are assessed. While finding the best panchromatic band, SWIR1 has been found to have a strong association (R2 = 0.512) with the average of four bands: Red, Green, Blue, and NIR. Similarly, VNIR and SWIR2 are found to be strongly correlated with the Red Band with R2 values of 0.68 and 0.43, respectively. High Pass Filter has the highest Universal Quality Index, lowest Spectral Angle Mapper, and lowest Root Mean Squared Error for pan-sharpening SWIR1, SWIR2, and VNIR bands. NDWI shows the highest Kappa (0.87) and F1-score (0.92). Following historical (2016–2023) analysis, Yen thresholding method shows an overall accuracy ranging from 82% in 2023 to 96% in 2021. Thus, the study suggests using Yen thresholding for NDWI in EH of Bangladesh.